Commit ·
af00fcc
1
Parent(s): 2d9333e
add new feature for enhanced model performance
Browse files- trainer_state.json +2433 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,2433 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.9152,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 300,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0064,
|
| 13 |
+
"grad_norm": 2.0177674293518066,
|
| 14 |
+
"learning_rate": 0.001,
|
| 15 |
+
"loss": 7.5774,
|
| 16 |
+
"mean_token_accuracy": 0.5871803313493729,
|
| 17 |
+
"step": 1
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.0128,
|
| 21 |
+
"grad_norm": 2.9027979373931885,
|
| 22 |
+
"learning_rate": 0.001,
|
| 23 |
+
"loss": 8.8307,
|
| 24 |
+
"mean_token_accuracy": 0.5205357447266579,
|
| 25 |
+
"step": 2
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 0.0192,
|
| 29 |
+
"grad_norm": 20.20247459411621,
|
| 30 |
+
"learning_rate": 0.001,
|
| 31 |
+
"loss": 10.8145,
|
| 32 |
+
"mean_token_accuracy": 0.46558595448732376,
|
| 33 |
+
"step": 3
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.0256,
|
| 37 |
+
"grad_norm": NaN,
|
| 38 |
+
"learning_rate": 0.001,
|
| 39 |
+
"loss": 12.6533,
|
| 40 |
+
"mean_token_accuracy": 0.4072755202651024,
|
| 41 |
+
"step": 4
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 0.032,
|
| 45 |
+
"grad_norm": NaN,
|
| 46 |
+
"learning_rate": 0.001,
|
| 47 |
+
"loss": 13.6561,
|
| 48 |
+
"mean_token_accuracy": 0.37730202078819275,
|
| 49 |
+
"step": 5
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 0.0384,
|
| 53 |
+
"grad_norm": 43.733421325683594,
|
| 54 |
+
"learning_rate": 0.001,
|
| 55 |
+
"loss": 12.4744,
|
| 56 |
+
"mean_token_accuracy": 0.40100549906492233,
|
| 57 |
+
"step": 6
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.0448,
|
| 61 |
+
"grad_norm": 19.141199111938477,
|
| 62 |
+
"learning_rate": 0.001,
|
| 63 |
+
"loss": 9.5344,
|
| 64 |
+
"mean_token_accuracy": 0.5192811861634254,
|
| 65 |
+
"step": 7
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.0512,
|
| 69 |
+
"grad_norm": 18.41023826599121,
|
| 70 |
+
"learning_rate": 0.001,
|
| 71 |
+
"loss": 14.3608,
|
| 72 |
+
"mean_token_accuracy": 0.43483711034059525,
|
| 73 |
+
"step": 8
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.0576,
|
| 77 |
+
"grad_norm": 43.658721923828125,
|
| 78 |
+
"learning_rate": 0.001,
|
| 79 |
+
"loss": 13.3078,
|
| 80 |
+
"mean_token_accuracy": 0.41958054155111313,
|
| 81 |
+
"step": 9
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.064,
|
| 85 |
+
"grad_norm": 8.837448120117188,
|
| 86 |
+
"learning_rate": 0.001,
|
| 87 |
+
"loss": 10.5372,
|
| 88 |
+
"mean_token_accuracy": 0.47571663558483124,
|
| 89 |
+
"step": 10
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.0704,
|
| 93 |
+
"grad_norm": 10.096052169799805,
|
| 94 |
+
"learning_rate": 0.001,
|
| 95 |
+
"loss": 12.1162,
|
| 96 |
+
"mean_token_accuracy": 0.4255446493625641,
|
| 97 |
+
"step": 11
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 0.0768,
|
| 101 |
+
"grad_norm": 5.100619316101074,
|
| 102 |
+
"learning_rate": 0.001,
|
| 103 |
+
"loss": 9.2243,
|
| 104 |
+
"mean_token_accuracy": 0.5292309895157814,
|
| 105 |
+
"step": 12
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.0832,
|
| 109 |
+
"grad_norm": 4.010665416717529,
|
| 110 |
+
"learning_rate": 0.001,
|
| 111 |
+
"loss": 6.7791,
|
| 112 |
+
"mean_token_accuracy": 0.6034770011901855,
|
| 113 |
+
"step": 13
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 0.0896,
|
| 117 |
+
"grad_norm": 7.71261739730835,
|
| 118 |
+
"learning_rate": 0.001,
|
| 119 |
+
"loss": 8.4717,
|
| 120 |
+
"mean_token_accuracy": 0.519000381231308,
|
| 121 |
+
"step": 14
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.096,
|
| 125 |
+
"grad_norm": 6.087244987487793,
|
| 126 |
+
"learning_rate": 0.001,
|
| 127 |
+
"loss": 8.6032,
|
| 128 |
+
"mean_token_accuracy": 0.5275572314858437,
|
| 129 |
+
"step": 15
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.1024,
|
| 133 |
+
"grad_norm": 3.9907681941986084,
|
| 134 |
+
"learning_rate": 0.001,
|
| 135 |
+
"loss": 7.6546,
|
| 136 |
+
"mean_token_accuracy": 0.542157307267189,
|
| 137 |
+
"step": 16
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.1088,
|
| 141 |
+
"grad_norm": 4.168624401092529,
|
| 142 |
+
"learning_rate": 0.001,
|
| 143 |
+
"loss": 7.6228,
|
| 144 |
+
"mean_token_accuracy": 0.5621763542294502,
|
| 145 |
+
"step": 17
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.1152,
|
| 149 |
+
"grad_norm": 4.9508490562438965,
|
| 150 |
+
"learning_rate": 0.001,
|
| 151 |
+
"loss": 9.808,
|
| 152 |
+
"mean_token_accuracy": 0.49128710478544235,
|
| 153 |
+
"step": 18
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.1216,
|
| 157 |
+
"grad_norm": 4.520578861236572,
|
| 158 |
+
"learning_rate": 0.001,
|
| 159 |
+
"loss": 8.8476,
|
| 160 |
+
"mean_token_accuracy": 0.4943431541323662,
|
| 161 |
+
"step": 19
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 0.128,
|
| 165 |
+
"grad_norm": 6.03663444519043,
|
| 166 |
+
"learning_rate": 0.001,
|
| 167 |
+
"loss": 8.0179,
|
| 168 |
+
"mean_token_accuracy": 0.5464938580989838,
|
| 169 |
+
"step": 20
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 0.1344,
|
| 173 |
+
"grad_norm": 4.153735637664795,
|
| 174 |
+
"learning_rate": 0.001,
|
| 175 |
+
"loss": 8.7043,
|
| 176 |
+
"mean_token_accuracy": 0.5475813895463943,
|
| 177 |
+
"step": 21
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.1408,
|
| 181 |
+
"grad_norm": 3.6078875064849854,
|
| 182 |
+
"learning_rate": 0.001,
|
| 183 |
+
"loss": 8.5597,
|
| 184 |
+
"mean_token_accuracy": 0.5447330102324486,
|
| 185 |
+
"step": 22
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.1472,
|
| 189 |
+
"grad_norm": 3.8478240966796875,
|
| 190 |
+
"learning_rate": 0.001,
|
| 191 |
+
"loss": 7.2146,
|
| 192 |
+
"mean_token_accuracy": 0.5992539674043655,
|
| 193 |
+
"step": 23
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.1536,
|
| 197 |
+
"grad_norm": 4.033797264099121,
|
| 198 |
+
"learning_rate": 0.001,
|
| 199 |
+
"loss": 9.3191,
|
| 200 |
+
"mean_token_accuracy": 0.5196451023221016,
|
| 201 |
+
"step": 24
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.16,
|
| 205 |
+
"grad_norm": 3.2458839416503906,
|
| 206 |
+
"learning_rate": 0.001,
|
| 207 |
+
"loss": 7.4417,
|
| 208 |
+
"mean_token_accuracy": 0.5597032383084297,
|
| 209 |
+
"step": 25
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 0.1664,
|
| 213 |
+
"grad_norm": 3.520216226577759,
|
| 214 |
+
"learning_rate": 0.001,
|
| 215 |
+
"loss": 6.7451,
|
| 216 |
+
"mean_token_accuracy": 0.5860115587711334,
|
| 217 |
+
"step": 26
|
| 218 |
+
},
|
| 219 |
+
{
|
| 220 |
+
"epoch": 0.1728,
|
| 221 |
+
"grad_norm": 3.7371273040771484,
|
| 222 |
+
"learning_rate": 0.001,
|
| 223 |
+
"loss": 7.8899,
|
| 224 |
+
"mean_token_accuracy": 0.5652190744876862,
|
| 225 |
+
"step": 27
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.1792,
|
| 229 |
+
"grad_norm": 4.060925006866455,
|
| 230 |
+
"learning_rate": 0.001,
|
| 231 |
+
"loss": 7.4009,
|
| 232 |
+
"mean_token_accuracy": 0.5925501734018326,
|
| 233 |
+
"step": 28
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.1856,
|
| 237 |
+
"grad_norm": 3.9062178134918213,
|
| 238 |
+
"learning_rate": 0.001,
|
| 239 |
+
"loss": 7.3145,
|
| 240 |
+
"mean_token_accuracy": 0.5906639099121094,
|
| 241 |
+
"step": 29
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.192,
|
| 245 |
+
"grad_norm": 3.7855491638183594,
|
| 246 |
+
"learning_rate": 0.001,
|
| 247 |
+
"loss": 7.9822,
|
| 248 |
+
"mean_token_accuracy": 0.5379130840301514,
|
| 249 |
+
"step": 30
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.1984,
|
| 253 |
+
"grad_norm": 3.340970516204834,
|
| 254 |
+
"learning_rate": 0.001,
|
| 255 |
+
"loss": 7.1766,
|
| 256 |
+
"mean_token_accuracy": 0.5813343971967697,
|
| 257 |
+
"step": 31
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 0.2048,
|
| 261 |
+
"grad_norm": 3.692194938659668,
|
| 262 |
+
"learning_rate": 0.001,
|
| 263 |
+
"loss": 7.6622,
|
| 264 |
+
"mean_token_accuracy": 0.5705007463693619,
|
| 265 |
+
"step": 32
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.2112,
|
| 269 |
+
"grad_norm": 3.3015754222869873,
|
| 270 |
+
"learning_rate": 0.001,
|
| 271 |
+
"loss": 7.1023,
|
| 272 |
+
"mean_token_accuracy": 0.5972073376178741,
|
| 273 |
+
"step": 33
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.2176,
|
| 277 |
+
"grad_norm": 3.191516160964966,
|
| 278 |
+
"learning_rate": 0.001,
|
| 279 |
+
"loss": 7.8106,
|
| 280 |
+
"mean_token_accuracy": 0.577879324555397,
|
| 281 |
+
"step": 34
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 0.224,
|
| 285 |
+
"grad_norm": 3.015953779220581,
|
| 286 |
+
"learning_rate": 0.001,
|
| 287 |
+
"loss": 8.2667,
|
| 288 |
+
"mean_token_accuracy": 0.5295384302735329,
|
| 289 |
+
"step": 35
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.2304,
|
| 293 |
+
"grad_norm": 3.0506298542022705,
|
| 294 |
+
"learning_rate": 0.001,
|
| 295 |
+
"loss": 8.2168,
|
| 296 |
+
"mean_token_accuracy": 0.5699651688337326,
|
| 297 |
+
"step": 36
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.2368,
|
| 301 |
+
"grad_norm": 3.11960506439209,
|
| 302 |
+
"learning_rate": 0.001,
|
| 303 |
+
"loss": 7.697,
|
| 304 |
+
"mean_token_accuracy": 0.5628390982747078,
|
| 305 |
+
"step": 37
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.2432,
|
| 309 |
+
"grad_norm": 3.0201468467712402,
|
| 310 |
+
"learning_rate": 0.001,
|
| 311 |
+
"loss": 7.6835,
|
| 312 |
+
"mean_token_accuracy": 0.5989912003278732,
|
| 313 |
+
"step": 38
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 0.2496,
|
| 317 |
+
"grad_norm": 3.462604522705078,
|
| 318 |
+
"learning_rate": 0.001,
|
| 319 |
+
"loss": 7.8899,
|
| 320 |
+
"mean_token_accuracy": 0.5595831423997879,
|
| 321 |
+
"step": 39
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.256,
|
| 325 |
+
"grad_norm": 3.401752233505249,
|
| 326 |
+
"learning_rate": 0.001,
|
| 327 |
+
"loss": 7.9503,
|
| 328 |
+
"mean_token_accuracy": 0.5780320316553116,
|
| 329 |
+
"step": 40
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 0.2624,
|
| 333 |
+
"grad_norm": 4.185473442077637,
|
| 334 |
+
"learning_rate": 0.001,
|
| 335 |
+
"loss": 7.799,
|
| 336 |
+
"mean_token_accuracy": 0.5548681393265724,
|
| 337 |
+
"step": 41
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 0.2688,
|
| 341 |
+
"grad_norm": 3.9332938194274902,
|
| 342 |
+
"learning_rate": 0.001,
|
| 343 |
+
"loss": 7.7152,
|
| 344 |
+
"mean_token_accuracy": 0.5636641383171082,
|
| 345 |
+
"step": 42
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.2752,
|
| 349 |
+
"grad_norm": 3.2568907737731934,
|
| 350 |
+
"learning_rate": 0.001,
|
| 351 |
+
"loss": 7.4339,
|
| 352 |
+
"mean_token_accuracy": 0.5839549452066422,
|
| 353 |
+
"step": 43
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.2816,
|
| 357 |
+
"grad_norm": 4.672369956970215,
|
| 358 |
+
"learning_rate": 0.001,
|
| 359 |
+
"loss": 9.5802,
|
| 360 |
+
"mean_token_accuracy": 0.5100178346037865,
|
| 361 |
+
"step": 44
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.288,
|
| 365 |
+
"grad_norm": 2.7107110023498535,
|
| 366 |
+
"learning_rate": 0.001,
|
| 367 |
+
"loss": 6.8152,
|
| 368 |
+
"mean_token_accuracy": 0.6055551916360855,
|
| 369 |
+
"step": 45
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.2944,
|
| 373 |
+
"grad_norm": 4.762430667877197,
|
| 374 |
+
"learning_rate": 0.001,
|
| 375 |
+
"loss": 9.5548,
|
| 376 |
+
"mean_token_accuracy": 0.5057197883725166,
|
| 377 |
+
"step": 46
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.3008,
|
| 381 |
+
"grad_norm": 3.7424633502960205,
|
| 382 |
+
"learning_rate": 0.001,
|
| 383 |
+
"loss": 8.5479,
|
| 384 |
+
"mean_token_accuracy": 0.5115512683987617,
|
| 385 |
+
"step": 47
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 0.3072,
|
| 389 |
+
"grad_norm": 3.2961721420288086,
|
| 390 |
+
"learning_rate": 0.001,
|
| 391 |
+
"loss": 8.2113,
|
| 392 |
+
"mean_token_accuracy": 0.557342454791069,
|
| 393 |
+
"step": 48
|
| 394 |
+
},
|
| 395 |
+
{
|
| 396 |
+
"epoch": 0.3136,
|
| 397 |
+
"grad_norm": 3.239579677581787,
|
| 398 |
+
"learning_rate": 0.001,
|
| 399 |
+
"loss": 8.8304,
|
| 400 |
+
"mean_token_accuracy": 0.5229588970541954,
|
| 401 |
+
"step": 49
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.32,
|
| 405 |
+
"grad_norm": 3.273914337158203,
|
| 406 |
+
"learning_rate": 0.001,
|
| 407 |
+
"loss": 7.7483,
|
| 408 |
+
"mean_token_accuracy": 0.5654094517230988,
|
| 409 |
+
"step": 50
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.3264,
|
| 413 |
+
"grad_norm": 3.6419663429260254,
|
| 414 |
+
"learning_rate": 0.001,
|
| 415 |
+
"loss": 7.3461,
|
| 416 |
+
"mean_token_accuracy": 0.6018975228071213,
|
| 417 |
+
"step": 51
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.3328,
|
| 421 |
+
"grad_norm": 3.4327375888824463,
|
| 422 |
+
"learning_rate": 0.001,
|
| 423 |
+
"loss": 7.6107,
|
| 424 |
+
"mean_token_accuracy": 0.5776806026697159,
|
| 425 |
+
"step": 52
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 0.3392,
|
| 429 |
+
"grad_norm": 2.8938372135162354,
|
| 430 |
+
"learning_rate": 0.001,
|
| 431 |
+
"loss": 7.7927,
|
| 432 |
+
"mean_token_accuracy": 0.5748834684491158,
|
| 433 |
+
"step": 53
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.3456,
|
| 437 |
+
"grad_norm": 3.780987024307251,
|
| 438 |
+
"learning_rate": 0.001,
|
| 439 |
+
"loss": 7.7148,
|
| 440 |
+
"mean_token_accuracy": 0.5800818204879761,
|
| 441 |
+
"step": 54
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.352,
|
| 445 |
+
"grad_norm": 7.018247604370117,
|
| 446 |
+
"learning_rate": 0.001,
|
| 447 |
+
"loss": 7.9333,
|
| 448 |
+
"mean_token_accuracy": 0.557623416185379,
|
| 449 |
+
"step": 55
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 0.3584,
|
| 453 |
+
"grad_norm": 4.366345405578613,
|
| 454 |
+
"learning_rate": 0.001,
|
| 455 |
+
"loss": 8.7168,
|
| 456 |
+
"mean_token_accuracy": 0.5106043368577957,
|
| 457 |
+
"step": 56
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.3648,
|
| 461 |
+
"grad_norm": 3.5044593811035156,
|
| 462 |
+
"learning_rate": 0.001,
|
| 463 |
+
"loss": 7.3359,
|
| 464 |
+
"mean_token_accuracy": 0.5744027197360992,
|
| 465 |
+
"step": 57
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.3712,
|
| 469 |
+
"grad_norm": 3.710293769836426,
|
| 470 |
+
"learning_rate": 0.001,
|
| 471 |
+
"loss": 7.1611,
|
| 472 |
+
"mean_token_accuracy": 0.6022080183029175,
|
| 473 |
+
"step": 58
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 0.3776,
|
| 477 |
+
"grad_norm": 3.784938097000122,
|
| 478 |
+
"learning_rate": 0.001,
|
| 479 |
+
"loss": 7.0631,
|
| 480 |
+
"mean_token_accuracy": 0.5821704566478729,
|
| 481 |
+
"step": 59
|
| 482 |
+
},
|
| 483 |
+
{
|
| 484 |
+
"epoch": 0.384,
|
| 485 |
+
"grad_norm": 3.79630708694458,
|
| 486 |
+
"learning_rate": 0.001,
|
| 487 |
+
"loss": 6.6392,
|
| 488 |
+
"mean_token_accuracy": 0.6132841110229492,
|
| 489 |
+
"step": 60
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.3904,
|
| 493 |
+
"grad_norm": 3.004554510116577,
|
| 494 |
+
"learning_rate": 0.001,
|
| 495 |
+
"loss": 7.1997,
|
| 496 |
+
"mean_token_accuracy": 0.5954362899065018,
|
| 497 |
+
"step": 61
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 0.3968,
|
| 501 |
+
"grad_norm": 4.015055179595947,
|
| 502 |
+
"learning_rate": 0.001,
|
| 503 |
+
"loss": 7.4768,
|
| 504 |
+
"mean_token_accuracy": 0.5736093372106552,
|
| 505 |
+
"step": 62
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.4032,
|
| 509 |
+
"grad_norm": 3.654245615005493,
|
| 510 |
+
"learning_rate": 0.001,
|
| 511 |
+
"loss": 7.8212,
|
| 512 |
+
"mean_token_accuracy": 0.5659709572792053,
|
| 513 |
+
"step": 63
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.4096,
|
| 517 |
+
"grad_norm": 4.0627641677856445,
|
| 518 |
+
"learning_rate": 0.001,
|
| 519 |
+
"loss": 9.2678,
|
| 520 |
+
"mean_token_accuracy": 0.5201347693800926,
|
| 521 |
+
"step": 64
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.416,
|
| 525 |
+
"grad_norm": 3.4690985679626465,
|
| 526 |
+
"learning_rate": 0.001,
|
| 527 |
+
"loss": 8.8517,
|
| 528 |
+
"mean_token_accuracy": 0.5343466401100159,
|
| 529 |
+
"step": 65
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 0.4224,
|
| 533 |
+
"grad_norm": 3.752857208251953,
|
| 534 |
+
"learning_rate": 0.001,
|
| 535 |
+
"loss": 8.8329,
|
| 536 |
+
"mean_token_accuracy": 0.5228310152888298,
|
| 537 |
+
"step": 66
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.4288,
|
| 541 |
+
"grad_norm": 2.968993902206421,
|
| 542 |
+
"learning_rate": 0.001,
|
| 543 |
+
"loss": 7.6058,
|
| 544 |
+
"mean_token_accuracy": 0.5344163477420807,
|
| 545 |
+
"step": 67
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 0.4352,
|
| 549 |
+
"grad_norm": 3.72929310798645,
|
| 550 |
+
"learning_rate": 0.001,
|
| 551 |
+
"loss": 9.3222,
|
| 552 |
+
"mean_token_accuracy": 0.516074076294899,
|
| 553 |
+
"step": 68
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 0.4416,
|
| 557 |
+
"grad_norm": 3.6030168533325195,
|
| 558 |
+
"learning_rate": 0.001,
|
| 559 |
+
"loss": 8.5513,
|
| 560 |
+
"mean_token_accuracy": 0.5421805381774902,
|
| 561 |
+
"step": 69
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.448,
|
| 565 |
+
"grad_norm": 3.1739273071289062,
|
| 566 |
+
"learning_rate": 0.001,
|
| 567 |
+
"loss": 7.0802,
|
| 568 |
+
"mean_token_accuracy": 0.5998114049434662,
|
| 569 |
+
"step": 70
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.4544,
|
| 573 |
+
"grad_norm": 2.880340099334717,
|
| 574 |
+
"learning_rate": 0.001,
|
| 575 |
+
"loss": 6.7046,
|
| 576 |
+
"mean_token_accuracy": 0.6100719571113586,
|
| 577 |
+
"step": 71
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.4608,
|
| 581 |
+
"grad_norm": 3.2981016635894775,
|
| 582 |
+
"learning_rate": 0.001,
|
| 583 |
+
"loss": 7.9998,
|
| 584 |
+
"mean_token_accuracy": 0.5657886415719986,
|
| 585 |
+
"step": 72
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.4672,
|
| 589 |
+
"grad_norm": 4.456881999969482,
|
| 590 |
+
"learning_rate": 0.001,
|
| 591 |
+
"loss": 7.1571,
|
| 592 |
+
"mean_token_accuracy": 0.5802148729562759,
|
| 593 |
+
"step": 73
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 0.4736,
|
| 597 |
+
"grad_norm": 4.069066047668457,
|
| 598 |
+
"learning_rate": 0.001,
|
| 599 |
+
"loss": 8.8054,
|
| 600 |
+
"mean_token_accuracy": 0.5348180085420609,
|
| 601 |
+
"step": 74
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 0.48,
|
| 605 |
+
"grad_norm": 9.803370475769043,
|
| 606 |
+
"learning_rate": 0.001,
|
| 607 |
+
"loss": 8.5662,
|
| 608 |
+
"mean_token_accuracy": 0.5325330048799515,
|
| 609 |
+
"step": 75
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.4864,
|
| 613 |
+
"grad_norm": 4.328041076660156,
|
| 614 |
+
"learning_rate": 0.001,
|
| 615 |
+
"loss": 8.1187,
|
| 616 |
+
"mean_token_accuracy": 0.5573177188634872,
|
| 617 |
+
"step": 76
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 0.4928,
|
| 621 |
+
"grad_norm": 5.3780975341796875,
|
| 622 |
+
"learning_rate": 0.001,
|
| 623 |
+
"loss": 8.1414,
|
| 624 |
+
"mean_token_accuracy": 0.5307755619287491,
|
| 625 |
+
"step": 77
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.4992,
|
| 629 |
+
"grad_norm": 4.601280212402344,
|
| 630 |
+
"learning_rate": 0.001,
|
| 631 |
+
"loss": 8.1343,
|
| 632 |
+
"mean_token_accuracy": 0.5730244666337967,
|
| 633 |
+
"step": 78
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.5056,
|
| 637 |
+
"grad_norm": 5.185049533843994,
|
| 638 |
+
"learning_rate": 0.001,
|
| 639 |
+
"loss": 10.3809,
|
| 640 |
+
"mean_token_accuracy": 0.45638662576675415,
|
| 641 |
+
"step": 79
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 0.512,
|
| 645 |
+
"grad_norm": 4.1511054039001465,
|
| 646 |
+
"learning_rate": 0.001,
|
| 647 |
+
"loss": 7.903,
|
| 648 |
+
"mean_token_accuracy": 0.5588131695985794,
|
| 649 |
+
"step": 80
|
| 650 |
+
},
|
| 651 |
+
{
|
| 652 |
+
"epoch": 0.5184,
|
| 653 |
+
"grad_norm": 3.8643414974212646,
|
| 654 |
+
"learning_rate": 0.001,
|
| 655 |
+
"loss": 9.1992,
|
| 656 |
+
"mean_token_accuracy": 0.526216022670269,
|
| 657 |
+
"step": 81
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.5248,
|
| 661 |
+
"grad_norm": 3.9921276569366455,
|
| 662 |
+
"learning_rate": 0.001,
|
| 663 |
+
"loss": 7.9791,
|
| 664 |
+
"mean_token_accuracy": 0.5335483700037003,
|
| 665 |
+
"step": 82
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 0.5312,
|
| 669 |
+
"grad_norm": 3.2328104972839355,
|
| 670 |
+
"learning_rate": 0.001,
|
| 671 |
+
"loss": 8.3164,
|
| 672 |
+
"mean_token_accuracy": 0.5265945047140121,
|
| 673 |
+
"step": 83
|
| 674 |
+
},
|
| 675 |
+
{
|
| 676 |
+
"epoch": 0.5376,
|
| 677 |
+
"grad_norm": 4.729416370391846,
|
| 678 |
+
"learning_rate": 0.001,
|
| 679 |
+
"loss": 8.9703,
|
| 680 |
+
"mean_token_accuracy": 0.5294319912791252,
|
| 681 |
+
"step": 84
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.544,
|
| 685 |
+
"grad_norm": 3.562079906463623,
|
| 686 |
+
"learning_rate": 0.001,
|
| 687 |
+
"loss": 8.2047,
|
| 688 |
+
"mean_token_accuracy": 0.557565338909626,
|
| 689 |
+
"step": 85
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.5504,
|
| 693 |
+
"grad_norm": 3.8010246753692627,
|
| 694 |
+
"learning_rate": 0.001,
|
| 695 |
+
"loss": 8.0639,
|
| 696 |
+
"mean_token_accuracy": 0.5428211987018585,
|
| 697 |
+
"step": 86
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 0.5568,
|
| 701 |
+
"grad_norm": 3.3087170124053955,
|
| 702 |
+
"learning_rate": 0.001,
|
| 703 |
+
"loss": 8.145,
|
| 704 |
+
"mean_token_accuracy": 0.5458494797348976,
|
| 705 |
+
"step": 87
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 0.5632,
|
| 709 |
+
"grad_norm": 3.956712245941162,
|
| 710 |
+
"learning_rate": 0.001,
|
| 711 |
+
"loss": 8.6958,
|
| 712 |
+
"mean_token_accuracy": 0.5472606420516968,
|
| 713 |
+
"step": 88
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 0.5696,
|
| 717 |
+
"grad_norm": 3.548285722732544,
|
| 718 |
+
"learning_rate": 0.001,
|
| 719 |
+
"loss": 7.0706,
|
| 720 |
+
"mean_token_accuracy": 0.605229914188385,
|
| 721 |
+
"step": 89
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 0.576,
|
| 725 |
+
"grad_norm": 3.6893436908721924,
|
| 726 |
+
"learning_rate": 0.001,
|
| 727 |
+
"loss": 6.2735,
|
| 728 |
+
"mean_token_accuracy": 0.6206476241350174,
|
| 729 |
+
"step": 90
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.5824,
|
| 733 |
+
"grad_norm": 8.876976013183594,
|
| 734 |
+
"learning_rate": 0.001,
|
| 735 |
+
"loss": 9.1893,
|
| 736 |
+
"mean_token_accuracy": 0.5234974399209023,
|
| 737 |
+
"step": 91
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.5888,
|
| 741 |
+
"grad_norm": 4.515777587890625,
|
| 742 |
+
"learning_rate": 0.001,
|
| 743 |
+
"loss": 7.1549,
|
| 744 |
+
"mean_token_accuracy": 0.5581575930118561,
|
| 745 |
+
"step": 92
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.5952,
|
| 749 |
+
"grad_norm": 3.814516067504883,
|
| 750 |
+
"learning_rate": 0.001,
|
| 751 |
+
"loss": 7.7558,
|
| 752 |
+
"mean_token_accuracy": 0.572398416697979,
|
| 753 |
+
"step": 93
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 0.6016,
|
| 757 |
+
"grad_norm": 4.637458801269531,
|
| 758 |
+
"learning_rate": 0.001,
|
| 759 |
+
"loss": 8.2918,
|
| 760 |
+
"mean_token_accuracy": 0.5480805337429047,
|
| 761 |
+
"step": 94
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 0.608,
|
| 765 |
+
"grad_norm": 3.827664613723755,
|
| 766 |
+
"learning_rate": 0.001,
|
| 767 |
+
"loss": 7.6153,
|
| 768 |
+
"mean_token_accuracy": 0.5540450513362885,
|
| 769 |
+
"step": 95
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 0.6144,
|
| 773 |
+
"grad_norm": 3.596336841583252,
|
| 774 |
+
"learning_rate": 0.001,
|
| 775 |
+
"loss": 8.1713,
|
| 776 |
+
"mean_token_accuracy": 0.5356313809752464,
|
| 777 |
+
"step": 96
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 0.6208,
|
| 781 |
+
"grad_norm": 8.448469161987305,
|
| 782 |
+
"learning_rate": 0.001,
|
| 783 |
+
"loss": 8.668,
|
| 784 |
+
"mean_token_accuracy": 0.5483140274882317,
|
| 785 |
+
"step": 97
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 0.6272,
|
| 789 |
+
"grad_norm": 5.3135576248168945,
|
| 790 |
+
"learning_rate": 0.001,
|
| 791 |
+
"loss": 7.7838,
|
| 792 |
+
"mean_token_accuracy": 0.5676964446902275,
|
| 793 |
+
"step": 98
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.6336,
|
| 797 |
+
"grad_norm": 4.343587398529053,
|
| 798 |
+
"learning_rate": 0.001,
|
| 799 |
+
"loss": 7.2509,
|
| 800 |
+
"mean_token_accuracy": 0.5730399489402771,
|
| 801 |
+
"step": 99
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.64,
|
| 805 |
+
"grad_norm": 3.066798686981201,
|
| 806 |
+
"learning_rate": 0.001,
|
| 807 |
+
"loss": 7.0591,
|
| 808 |
+
"mean_token_accuracy": 0.6044407039880753,
|
| 809 |
+
"step": 100
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 0.6464,
|
| 813 |
+
"grad_norm": 4.594414234161377,
|
| 814 |
+
"learning_rate": 0.001,
|
| 815 |
+
"loss": 7.8146,
|
| 816 |
+
"mean_token_accuracy": 0.5625345483422279,
|
| 817 |
+
"step": 101
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 0.6528,
|
| 821 |
+
"grad_norm": 7.372500419616699,
|
| 822 |
+
"learning_rate": 0.001,
|
| 823 |
+
"loss": 7.9078,
|
| 824 |
+
"mean_token_accuracy": 0.5716247260570526,
|
| 825 |
+
"step": 102
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 0.6592,
|
| 829 |
+
"grad_norm": 4.589946269989014,
|
| 830 |
+
"learning_rate": 0.001,
|
| 831 |
+
"loss": 7.0762,
|
| 832 |
+
"mean_token_accuracy": 0.5866946280002594,
|
| 833 |
+
"step": 103
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 0.6656,
|
| 837 |
+
"grad_norm": 8.939701080322266,
|
| 838 |
+
"learning_rate": 0.001,
|
| 839 |
+
"loss": 8.9255,
|
| 840 |
+
"mean_token_accuracy": 0.528026632964611,
|
| 841 |
+
"step": 104
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 0.672,
|
| 845 |
+
"grad_norm": 4.3290534019470215,
|
| 846 |
+
"learning_rate": 0.001,
|
| 847 |
+
"loss": 8.1217,
|
| 848 |
+
"mean_token_accuracy": 0.5670952945947647,
|
| 849 |
+
"step": 105
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.6784,
|
| 853 |
+
"grad_norm": 3.7743752002716064,
|
| 854 |
+
"learning_rate": 0.001,
|
| 855 |
+
"loss": 7.4548,
|
| 856 |
+
"mean_token_accuracy": 0.5619177967309952,
|
| 857 |
+
"step": 106
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.6848,
|
| 861 |
+
"grad_norm": 4.942656993865967,
|
| 862 |
+
"learning_rate": 0.001,
|
| 863 |
+
"loss": 8.1344,
|
| 864 |
+
"mean_token_accuracy": 0.565968431532383,
|
| 865 |
+
"step": 107
|
| 866 |
+
},
|
| 867 |
+
{
|
| 868 |
+
"epoch": 0.6912,
|
| 869 |
+
"grad_norm": 4.2500715255737305,
|
| 870 |
+
"learning_rate": 0.001,
|
| 871 |
+
"loss": 7.5658,
|
| 872 |
+
"mean_token_accuracy": 0.5492838993668556,
|
| 873 |
+
"step": 108
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.6976,
|
| 877 |
+
"grad_norm": 3.6989309787750244,
|
| 878 |
+
"learning_rate": 0.001,
|
| 879 |
+
"loss": 7.7848,
|
| 880 |
+
"mean_token_accuracy": 0.5549194514751434,
|
| 881 |
+
"step": 109
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 0.704,
|
| 885 |
+
"grad_norm": 4.687801837921143,
|
| 886 |
+
"learning_rate": 0.001,
|
| 887 |
+
"loss": 8.1714,
|
| 888 |
+
"mean_token_accuracy": 0.5541109219193459,
|
| 889 |
+
"step": 110
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 0.7104,
|
| 893 |
+
"grad_norm": 3.4812569618225098,
|
| 894 |
+
"learning_rate": 0.001,
|
| 895 |
+
"loss": 9.1957,
|
| 896 |
+
"mean_token_accuracy": 0.5032462030649185,
|
| 897 |
+
"step": 111
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 0.7168,
|
| 901 |
+
"grad_norm": 4.627752780914307,
|
| 902 |
+
"learning_rate": 0.001,
|
| 903 |
+
"loss": 9.3127,
|
| 904 |
+
"mean_token_accuracy": 0.5073471590876579,
|
| 905 |
+
"step": 112
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.7232,
|
| 909 |
+
"grad_norm": 3.9376790523529053,
|
| 910 |
+
"learning_rate": 0.001,
|
| 911 |
+
"loss": 8.562,
|
| 912 |
+
"mean_token_accuracy": 0.5392563939094543,
|
| 913 |
+
"step": 113
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.7296,
|
| 917 |
+
"grad_norm": 3.162787675857544,
|
| 918 |
+
"learning_rate": 0.001,
|
| 919 |
+
"loss": 7.3301,
|
| 920 |
+
"mean_token_accuracy": 0.5914620906114578,
|
| 921 |
+
"step": 114
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 0.736,
|
| 925 |
+
"grad_norm": 4.065870761871338,
|
| 926 |
+
"learning_rate": 0.001,
|
| 927 |
+
"loss": 7.881,
|
| 928 |
+
"mean_token_accuracy": 0.5708122700452805,
|
| 929 |
+
"step": 115
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 0.7424,
|
| 933 |
+
"grad_norm": 4.9694318771362305,
|
| 934 |
+
"learning_rate": 0.001,
|
| 935 |
+
"loss": 7.6188,
|
| 936 |
+
"mean_token_accuracy": 0.5728907287120819,
|
| 937 |
+
"step": 116
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 0.7488,
|
| 941 |
+
"grad_norm": 3.600388526916504,
|
| 942 |
+
"learning_rate": 0.001,
|
| 943 |
+
"loss": 6.7958,
|
| 944 |
+
"mean_token_accuracy": 0.6140304803848267,
|
| 945 |
+
"step": 117
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 0.7552,
|
| 949 |
+
"grad_norm": 4.772489070892334,
|
| 950 |
+
"learning_rate": 0.001,
|
| 951 |
+
"loss": 7.8477,
|
| 952 |
+
"mean_token_accuracy": 0.5720326602458954,
|
| 953 |
+
"step": 118
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 0.7616,
|
| 957 |
+
"grad_norm": 4.109669208526611,
|
| 958 |
+
"learning_rate": 0.001,
|
| 959 |
+
"loss": 7.8857,
|
| 960 |
+
"mean_token_accuracy": 0.5725124627351761,
|
| 961 |
+
"step": 119
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.768,
|
| 965 |
+
"grad_norm": 3.9959371089935303,
|
| 966 |
+
"learning_rate": 0.001,
|
| 967 |
+
"loss": 7.8924,
|
| 968 |
+
"mean_token_accuracy": 0.5729866176843643,
|
| 969 |
+
"step": 120
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.7744,
|
| 973 |
+
"grad_norm": 4.218033790588379,
|
| 974 |
+
"learning_rate": 0.001,
|
| 975 |
+
"loss": 8.5045,
|
| 976 |
+
"mean_token_accuracy": 0.5448136478662491,
|
| 977 |
+
"step": 121
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 0.7808,
|
| 981 |
+
"grad_norm": 10.680689811706543,
|
| 982 |
+
"learning_rate": 0.001,
|
| 983 |
+
"loss": 8.5677,
|
| 984 |
+
"mean_token_accuracy": 0.5437836945056915,
|
| 985 |
+
"step": 122
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 0.7872,
|
| 989 |
+
"grad_norm": 6.1264729499816895,
|
| 990 |
+
"learning_rate": 0.001,
|
| 991 |
+
"loss": 10.1013,
|
| 992 |
+
"mean_token_accuracy": 0.4813794791698456,
|
| 993 |
+
"step": 123
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 0.7936,
|
| 997 |
+
"grad_norm": 3.4104678630828857,
|
| 998 |
+
"learning_rate": 0.001,
|
| 999 |
+
"loss": 7.5739,
|
| 1000 |
+
"mean_token_accuracy": 0.5576579570770264,
|
| 1001 |
+
"step": 124
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 0.8,
|
| 1005 |
+
"grad_norm": 4.8629255294799805,
|
| 1006 |
+
"learning_rate": 0.001,
|
| 1007 |
+
"loss": 8.2406,
|
| 1008 |
+
"mean_token_accuracy": 0.5386597216129303,
|
| 1009 |
+
"step": 125
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 0.8064,
|
| 1013 |
+
"grad_norm": 4.2899627685546875,
|
| 1014 |
+
"learning_rate": 0.001,
|
| 1015 |
+
"loss": 8.4591,
|
| 1016 |
+
"mean_token_accuracy": 0.5415358692407608,
|
| 1017 |
+
"step": 126
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.8128,
|
| 1021 |
+
"grad_norm": 3.7481234073638916,
|
| 1022 |
+
"learning_rate": 0.001,
|
| 1023 |
+
"loss": 7.0793,
|
| 1024 |
+
"mean_token_accuracy": 0.6038873642683029,
|
| 1025 |
+
"step": 127
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.8192,
|
| 1029 |
+
"grad_norm": 3.3412671089172363,
|
| 1030 |
+
"learning_rate": 0.001,
|
| 1031 |
+
"loss": 6.6192,
|
| 1032 |
+
"mean_token_accuracy": 0.6217593550682068,
|
| 1033 |
+
"step": 128
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"epoch": 0.8256,
|
| 1037 |
+
"grad_norm": 4.920543670654297,
|
| 1038 |
+
"learning_rate": 0.001,
|
| 1039 |
+
"loss": 8.4534,
|
| 1040 |
+
"mean_token_accuracy": 0.5469236895442009,
|
| 1041 |
+
"step": 129
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 0.832,
|
| 1045 |
+
"grad_norm": 3.746335029602051,
|
| 1046 |
+
"learning_rate": 0.001,
|
| 1047 |
+
"loss": 8.2707,
|
| 1048 |
+
"mean_token_accuracy": 0.5559498444199562,
|
| 1049 |
+
"step": 130
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 0.8384,
|
| 1053 |
+
"grad_norm": 3.816923141479492,
|
| 1054 |
+
"learning_rate": 0.001,
|
| 1055 |
+
"loss": 7.3012,
|
| 1056 |
+
"mean_token_accuracy": 0.5551306903362274,
|
| 1057 |
+
"step": 131
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 0.8448,
|
| 1061 |
+
"grad_norm": 5.7253265380859375,
|
| 1062 |
+
"learning_rate": 0.001,
|
| 1063 |
+
"loss": 7.7264,
|
| 1064 |
+
"mean_token_accuracy": 0.5539910942316055,
|
| 1065 |
+
"step": 132
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 0.8512,
|
| 1069 |
+
"grad_norm": 9.863517761230469,
|
| 1070 |
+
"learning_rate": 0.001,
|
| 1071 |
+
"loss": 8.4424,
|
| 1072 |
+
"mean_token_accuracy": 0.551963597536087,
|
| 1073 |
+
"step": 133
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.8576,
|
| 1077 |
+
"grad_norm": 5.7852582931518555,
|
| 1078 |
+
"learning_rate": 0.001,
|
| 1079 |
+
"loss": 8.8394,
|
| 1080 |
+
"mean_token_accuracy": 0.503598153591156,
|
| 1081 |
+
"step": 134
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.864,
|
| 1085 |
+
"grad_norm": 4.214358806610107,
|
| 1086 |
+
"learning_rate": 0.001,
|
| 1087 |
+
"loss": 8.0501,
|
| 1088 |
+
"mean_token_accuracy": 0.578537181019783,
|
| 1089 |
+
"step": 135
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 0.8704,
|
| 1093 |
+
"grad_norm": 4.677027225494385,
|
| 1094 |
+
"learning_rate": 0.001,
|
| 1095 |
+
"loss": 8.2782,
|
| 1096 |
+
"mean_token_accuracy": 0.556526243686676,
|
| 1097 |
+
"step": 136
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 0.8768,
|
| 1101 |
+
"grad_norm": 4.3732590675354,
|
| 1102 |
+
"learning_rate": 0.001,
|
| 1103 |
+
"loss": 7.1854,
|
| 1104 |
+
"mean_token_accuracy": 0.5627769976854324,
|
| 1105 |
+
"step": 137
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 0.8832,
|
| 1109 |
+
"grad_norm": 4.285269260406494,
|
| 1110 |
+
"learning_rate": 0.001,
|
| 1111 |
+
"loss": 7.5983,
|
| 1112 |
+
"mean_token_accuracy": 0.5761594474315643,
|
| 1113 |
+
"step": 138
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 0.8896,
|
| 1117 |
+
"grad_norm": 4.189344882965088,
|
| 1118 |
+
"learning_rate": 0.001,
|
| 1119 |
+
"loss": 7.796,
|
| 1120 |
+
"mean_token_accuracy": 0.5710103362798691,
|
| 1121 |
+
"step": 139
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 0.896,
|
| 1125 |
+
"grad_norm": 5.359838008880615,
|
| 1126 |
+
"learning_rate": 0.001,
|
| 1127 |
+
"loss": 9.3741,
|
| 1128 |
+
"mean_token_accuracy": 0.522984579205513,
|
| 1129 |
+
"step": 140
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.9024,
|
| 1133 |
+
"grad_norm": 4.133533000946045,
|
| 1134 |
+
"learning_rate": 0.001,
|
| 1135 |
+
"loss": 8.7714,
|
| 1136 |
+
"mean_token_accuracy": 0.49868279695510864,
|
| 1137 |
+
"step": 141
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.9088,
|
| 1141 |
+
"grad_norm": 3.432337522506714,
|
| 1142 |
+
"learning_rate": 0.001,
|
| 1143 |
+
"loss": 7.1933,
|
| 1144 |
+
"mean_token_accuracy": 0.5911879688501358,
|
| 1145 |
+
"step": 142
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 0.9152,
|
| 1149 |
+
"grad_norm": 3.4508094787597656,
|
| 1150 |
+
"learning_rate": 0.001,
|
| 1151 |
+
"loss": 8.4234,
|
| 1152 |
+
"mean_token_accuracy": 0.544566810131073,
|
| 1153 |
+
"step": 143
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 0.9216,
|
| 1157 |
+
"grad_norm": 4.863613605499268,
|
| 1158 |
+
"learning_rate": 0.001,
|
| 1159 |
+
"loss": 8.2328,
|
| 1160 |
+
"mean_token_accuracy": 0.5418634340167046,
|
| 1161 |
+
"step": 144
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 0.928,
|
| 1165 |
+
"grad_norm": 3.8848912715911865,
|
| 1166 |
+
"learning_rate": 0.001,
|
| 1167 |
+
"loss": 8.2236,
|
| 1168 |
+
"mean_token_accuracy": 0.5698046088218689,
|
| 1169 |
+
"step": 145
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 0.9344,
|
| 1173 |
+
"grad_norm": NaN,
|
| 1174 |
+
"learning_rate": 0.001,
|
| 1175 |
+
"loss": 8.5748,
|
| 1176 |
+
"mean_token_accuracy": 0.5264320075511932,
|
| 1177 |
+
"step": 146
|
| 1178 |
+
},
|
| 1179 |
+
{
|
| 1180 |
+
"epoch": 0.9408,
|
| 1181 |
+
"grad_norm": 4.209794998168945,
|
| 1182 |
+
"learning_rate": 0.001,
|
| 1183 |
+
"loss": 8.6393,
|
| 1184 |
+
"mean_token_accuracy": 0.5443025454878807,
|
| 1185 |
+
"step": 147
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.9472,
|
| 1189 |
+
"grad_norm": 5.81450891494751,
|
| 1190 |
+
"learning_rate": 0.001,
|
| 1191 |
+
"loss": 8.0271,
|
| 1192 |
+
"mean_token_accuracy": 0.5531649142503738,
|
| 1193 |
+
"step": 148
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.9536,
|
| 1197 |
+
"grad_norm": 3.847665786743164,
|
| 1198 |
+
"learning_rate": 0.001,
|
| 1199 |
+
"loss": 7.4199,
|
| 1200 |
+
"mean_token_accuracy": 0.5528692901134491,
|
| 1201 |
+
"step": 149
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 0.96,
|
| 1205 |
+
"grad_norm": 4.067963123321533,
|
| 1206 |
+
"learning_rate": 0.001,
|
| 1207 |
+
"loss": 6.8948,
|
| 1208 |
+
"mean_token_accuracy": 0.6041955649852753,
|
| 1209 |
+
"step": 150
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.9664,
|
| 1213 |
+
"grad_norm": 3.6745011806488037,
|
| 1214 |
+
"learning_rate": 0.001,
|
| 1215 |
+
"loss": 7.916,
|
| 1216 |
+
"mean_token_accuracy": 0.5488733500242233,
|
| 1217 |
+
"step": 151
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 0.9728,
|
| 1221 |
+
"grad_norm": 3.931896686553955,
|
| 1222 |
+
"learning_rate": 0.001,
|
| 1223 |
+
"loss": 7.7169,
|
| 1224 |
+
"mean_token_accuracy": 0.5718288719654083,
|
| 1225 |
+
"step": 152
|
| 1226 |
+
},
|
| 1227 |
+
{
|
| 1228 |
+
"epoch": 0.9792,
|
| 1229 |
+
"grad_norm": 4.551197052001953,
|
| 1230 |
+
"learning_rate": 0.001,
|
| 1231 |
+
"loss": 8.1617,
|
| 1232 |
+
"mean_token_accuracy": 0.5373415797948837,
|
| 1233 |
+
"step": 153
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 0.9856,
|
| 1237 |
+
"grad_norm": 4.994777679443359,
|
| 1238 |
+
"learning_rate": 0.001,
|
| 1239 |
+
"loss": 6.9665,
|
| 1240 |
+
"mean_token_accuracy": 0.5668836236000061,
|
| 1241 |
+
"step": 154
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.992,
|
| 1245 |
+
"grad_norm": 4.934599876403809,
|
| 1246 |
+
"learning_rate": 0.001,
|
| 1247 |
+
"loss": 8.9109,
|
| 1248 |
+
"mean_token_accuracy": 0.5285608246922493,
|
| 1249 |
+
"step": 155
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 0.9984,
|
| 1253 |
+
"grad_norm": 4.362490653991699,
|
| 1254 |
+
"learning_rate": 0.001,
|
| 1255 |
+
"loss": 7.9886,
|
| 1256 |
+
"mean_token_accuracy": 0.5652178972959518,
|
| 1257 |
+
"step": 156
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 1.0,
|
| 1261 |
+
"grad_norm": 2.0191662311553955,
|
| 1262 |
+
"learning_rate": 0.001,
|
| 1263 |
+
"loss": 2.2094,
|
| 1264 |
+
"mean_token_accuracy": 0.49253731966018677,
|
| 1265 |
+
"step": 157
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 1.0064,
|
| 1269 |
+
"grad_norm": 4.179709434509277,
|
| 1270 |
+
"learning_rate": 0.001,
|
| 1271 |
+
"loss": 6.0564,
|
| 1272 |
+
"mean_token_accuracy": 0.6441845297813416,
|
| 1273 |
+
"step": 158
|
| 1274 |
+
},
|
| 1275 |
+
{
|
| 1276 |
+
"epoch": 1.0128,
|
| 1277 |
+
"grad_norm": 4.1790971755981445,
|
| 1278 |
+
"learning_rate": 0.001,
|
| 1279 |
+
"loss": 7.223,
|
| 1280 |
+
"mean_token_accuracy": 0.5850173979997635,
|
| 1281 |
+
"step": 159
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 1.0192,
|
| 1285 |
+
"grad_norm": 3.393885374069214,
|
| 1286 |
+
"learning_rate": 0.001,
|
| 1287 |
+
"loss": 5.8742,
|
| 1288 |
+
"mean_token_accuracy": 0.6555072665214539,
|
| 1289 |
+
"step": 160
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 1.0256,
|
| 1293 |
+
"grad_norm": 4.170093536376953,
|
| 1294 |
+
"learning_rate": 0.001,
|
| 1295 |
+
"loss": 6.4492,
|
| 1296 |
+
"mean_token_accuracy": 0.6223380863666534,
|
| 1297 |
+
"step": 161
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 1.032,
|
| 1301 |
+
"grad_norm": 4.816079616546631,
|
| 1302 |
+
"learning_rate": 0.001,
|
| 1303 |
+
"loss": 8.458,
|
| 1304 |
+
"mean_token_accuracy": 0.5561687871813774,
|
| 1305 |
+
"step": 162
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 1.0384,
|
| 1309 |
+
"grad_norm": 7.0680108070373535,
|
| 1310 |
+
"learning_rate": 0.001,
|
| 1311 |
+
"loss": 6.4478,
|
| 1312 |
+
"mean_token_accuracy": 0.6458811461925507,
|
| 1313 |
+
"step": 163
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 1.0448,
|
| 1317 |
+
"grad_norm": 5.175418376922607,
|
| 1318 |
+
"learning_rate": 0.001,
|
| 1319 |
+
"loss": 8.6144,
|
| 1320 |
+
"mean_token_accuracy": 0.5237443894147873,
|
| 1321 |
+
"step": 164
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 1.0512,
|
| 1325 |
+
"grad_norm": 3.455251693725586,
|
| 1326 |
+
"learning_rate": 0.001,
|
| 1327 |
+
"loss": 6.6271,
|
| 1328 |
+
"mean_token_accuracy": 0.6214667707681656,
|
| 1329 |
+
"step": 165
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 1.0576,
|
| 1333 |
+
"grad_norm": 4.4586992263793945,
|
| 1334 |
+
"learning_rate": 0.001,
|
| 1335 |
+
"loss": 5.9971,
|
| 1336 |
+
"mean_token_accuracy": 0.654425784945488,
|
| 1337 |
+
"step": 166
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 1.064,
|
| 1341 |
+
"grad_norm": 4.797828197479248,
|
| 1342 |
+
"learning_rate": 0.001,
|
| 1343 |
+
"loss": 7.2682,
|
| 1344 |
+
"mean_token_accuracy": 0.5919231176376343,
|
| 1345 |
+
"step": 167
|
| 1346 |
+
},
|
| 1347 |
+
{
|
| 1348 |
+
"epoch": 1.0704,
|
| 1349 |
+
"grad_norm": 4.61244010925293,
|
| 1350 |
+
"learning_rate": 0.001,
|
| 1351 |
+
"loss": 6.1862,
|
| 1352 |
+
"mean_token_accuracy": 0.6452524065971375,
|
| 1353 |
+
"step": 168
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 1.0768,
|
| 1357 |
+
"grad_norm": 5.201780796051025,
|
| 1358 |
+
"learning_rate": 0.001,
|
| 1359 |
+
"loss": 9.1169,
|
| 1360 |
+
"mean_token_accuracy": 0.5475455224514008,
|
| 1361 |
+
"step": 169
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 1.0832,
|
| 1365 |
+
"grad_norm": 5.166975975036621,
|
| 1366 |
+
"learning_rate": 0.001,
|
| 1367 |
+
"loss": 6.9593,
|
| 1368 |
+
"mean_token_accuracy": 0.5929352939128876,
|
| 1369 |
+
"step": 170
|
| 1370 |
+
},
|
| 1371 |
+
{
|
| 1372 |
+
"epoch": 1.0896,
|
| 1373 |
+
"grad_norm": 5.824632167816162,
|
| 1374 |
+
"learning_rate": 0.001,
|
| 1375 |
+
"loss": 8.294,
|
| 1376 |
+
"mean_token_accuracy": 0.5294871926307678,
|
| 1377 |
+
"step": 171
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 1.096,
|
| 1381 |
+
"grad_norm": 9.257010459899902,
|
| 1382 |
+
"learning_rate": 0.001,
|
| 1383 |
+
"loss": 7.0368,
|
| 1384 |
+
"mean_token_accuracy": 0.6035382002592087,
|
| 1385 |
+
"step": 172
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 1.1024,
|
| 1389 |
+
"grad_norm": 5.524269104003906,
|
| 1390 |
+
"learning_rate": 0.001,
|
| 1391 |
+
"loss": 5.6529,
|
| 1392 |
+
"mean_token_accuracy": 0.6525093466043472,
|
| 1393 |
+
"step": 173
|
| 1394 |
+
},
|
| 1395 |
+
{
|
| 1396 |
+
"epoch": 1.1088,
|
| 1397 |
+
"grad_norm": 5.32748556137085,
|
| 1398 |
+
"learning_rate": 0.001,
|
| 1399 |
+
"loss": 6.7001,
|
| 1400 |
+
"mean_token_accuracy": 0.6103013008832932,
|
| 1401 |
+
"step": 174
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 1.1152,
|
| 1405 |
+
"grad_norm": 5.178861141204834,
|
| 1406 |
+
"learning_rate": 0.001,
|
| 1407 |
+
"loss": 6.7637,
|
| 1408 |
+
"mean_token_accuracy": 0.6111182868480682,
|
| 1409 |
+
"step": 175
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 1.1216,
|
| 1413 |
+
"grad_norm": 3.9681689739227295,
|
| 1414 |
+
"learning_rate": 0.001,
|
| 1415 |
+
"loss": 6.0235,
|
| 1416 |
+
"mean_token_accuracy": 0.6353351175785065,
|
| 1417 |
+
"step": 176
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 1.1280000000000001,
|
| 1421 |
+
"grad_norm": 4.23559045791626,
|
| 1422 |
+
"learning_rate": 0.001,
|
| 1423 |
+
"loss": 6.2243,
|
| 1424 |
+
"mean_token_accuracy": 0.636825680732727,
|
| 1425 |
+
"step": 177
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 1.1344,
|
| 1429 |
+
"grad_norm": 9.25437068939209,
|
| 1430 |
+
"learning_rate": 0.001,
|
| 1431 |
+
"loss": 7.3467,
|
| 1432 |
+
"mean_token_accuracy": 0.6091018617153168,
|
| 1433 |
+
"step": 178
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 1.1408,
|
| 1437 |
+
"grad_norm": 6.518167018890381,
|
| 1438 |
+
"learning_rate": 0.001,
|
| 1439 |
+
"loss": 7.4647,
|
| 1440 |
+
"mean_token_accuracy": 0.5769879817962646,
|
| 1441 |
+
"step": 179
|
| 1442 |
+
},
|
| 1443 |
+
{
|
| 1444 |
+
"epoch": 1.1472,
|
| 1445 |
+
"grad_norm": 8.98256778717041,
|
| 1446 |
+
"learning_rate": 0.001,
|
| 1447 |
+
"loss": 8.0621,
|
| 1448 |
+
"mean_token_accuracy": 0.5615440681576729,
|
| 1449 |
+
"step": 180
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 1.1536,
|
| 1453 |
+
"grad_norm": 6.962597370147705,
|
| 1454 |
+
"learning_rate": 0.001,
|
| 1455 |
+
"loss": 7.0736,
|
| 1456 |
+
"mean_token_accuracy": 0.6006975322961807,
|
| 1457 |
+
"step": 181
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 1.16,
|
| 1461 |
+
"grad_norm": 15.468770027160645,
|
| 1462 |
+
"learning_rate": 0.001,
|
| 1463 |
+
"loss": 7.8873,
|
| 1464 |
+
"mean_token_accuracy": 0.5628447085618973,
|
| 1465 |
+
"step": 182
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 1.1663999999999999,
|
| 1469 |
+
"grad_norm": 6.4656596183776855,
|
| 1470 |
+
"learning_rate": 0.001,
|
| 1471 |
+
"loss": 7.7006,
|
| 1472 |
+
"mean_token_accuracy": 0.5857141762971878,
|
| 1473 |
+
"step": 183
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 1.1728,
|
| 1477 |
+
"grad_norm": 6.264251708984375,
|
| 1478 |
+
"learning_rate": 0.001,
|
| 1479 |
+
"loss": 8.5055,
|
| 1480 |
+
"mean_token_accuracy": 0.5526725947856903,
|
| 1481 |
+
"step": 184
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 1.1792,
|
| 1485 |
+
"grad_norm": 9.173776626586914,
|
| 1486 |
+
"learning_rate": 0.001,
|
| 1487 |
+
"loss": 8.3511,
|
| 1488 |
+
"mean_token_accuracy": 0.5329839885234833,
|
| 1489 |
+
"step": 185
|
| 1490 |
+
},
|
| 1491 |
+
{
|
| 1492 |
+
"epoch": 1.1856,
|
| 1493 |
+
"grad_norm": 6.497045993804932,
|
| 1494 |
+
"learning_rate": 0.001,
|
| 1495 |
+
"loss": 8.5874,
|
| 1496 |
+
"mean_token_accuracy": 0.5194432288408279,
|
| 1497 |
+
"step": 186
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 1.192,
|
| 1501 |
+
"grad_norm": 6.55042839050293,
|
| 1502 |
+
"learning_rate": 0.001,
|
| 1503 |
+
"loss": 7.2464,
|
| 1504 |
+
"mean_token_accuracy": 0.5852644890546799,
|
| 1505 |
+
"step": 187
|
| 1506 |
+
},
|
| 1507 |
+
{
|
| 1508 |
+
"epoch": 1.1984,
|
| 1509 |
+
"grad_norm": 4.928357124328613,
|
| 1510 |
+
"learning_rate": 0.001,
|
| 1511 |
+
"loss": 7.421,
|
| 1512 |
+
"mean_token_accuracy": 0.575689509510994,
|
| 1513 |
+
"step": 188
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 1.2048,
|
| 1517 |
+
"grad_norm": 6.555416107177734,
|
| 1518 |
+
"learning_rate": 0.001,
|
| 1519 |
+
"loss": 7.1958,
|
| 1520 |
+
"mean_token_accuracy": 0.6004299223423004,
|
| 1521 |
+
"step": 189
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 1.2112,
|
| 1525 |
+
"grad_norm": 6.944830417633057,
|
| 1526 |
+
"learning_rate": 0.001,
|
| 1527 |
+
"loss": 6.8997,
|
| 1528 |
+
"mean_token_accuracy": 0.5822109580039978,
|
| 1529 |
+
"step": 190
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 1.2176,
|
| 1533 |
+
"grad_norm": 8.618417739868164,
|
| 1534 |
+
"learning_rate": 0.001,
|
| 1535 |
+
"loss": 8.1635,
|
| 1536 |
+
"mean_token_accuracy": 0.5582073777914047,
|
| 1537 |
+
"step": 191
|
| 1538 |
+
},
|
| 1539 |
+
{
|
| 1540 |
+
"epoch": 1.224,
|
| 1541 |
+
"grad_norm": 8.295172691345215,
|
| 1542 |
+
"learning_rate": 0.001,
|
| 1543 |
+
"loss": 7.5801,
|
| 1544 |
+
"mean_token_accuracy": 0.5919029116630554,
|
| 1545 |
+
"step": 192
|
| 1546 |
+
},
|
| 1547 |
+
{
|
| 1548 |
+
"epoch": 1.2304,
|
| 1549 |
+
"grad_norm": 12.239678382873535,
|
| 1550 |
+
"learning_rate": 0.001,
|
| 1551 |
+
"loss": 7.3438,
|
| 1552 |
+
"mean_token_accuracy": 0.5995124354958534,
|
| 1553 |
+
"step": 193
|
| 1554 |
+
},
|
| 1555 |
+
{
|
| 1556 |
+
"epoch": 1.2368000000000001,
|
| 1557 |
+
"grad_norm": 7.0297441482543945,
|
| 1558 |
+
"learning_rate": 0.001,
|
| 1559 |
+
"loss": 8.2068,
|
| 1560 |
+
"mean_token_accuracy": 0.537700891494751,
|
| 1561 |
+
"step": 194
|
| 1562 |
+
},
|
| 1563 |
+
{
|
| 1564 |
+
"epoch": 1.2432,
|
| 1565 |
+
"grad_norm": 5.0657639503479,
|
| 1566 |
+
"learning_rate": 0.001,
|
| 1567 |
+
"loss": 7.3465,
|
| 1568 |
+
"mean_token_accuracy": 0.587388165295124,
|
| 1569 |
+
"step": 195
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 1.2496,
|
| 1573 |
+
"grad_norm": 5.155961990356445,
|
| 1574 |
+
"learning_rate": 0.001,
|
| 1575 |
+
"loss": 7.4988,
|
| 1576 |
+
"mean_token_accuracy": 0.6019035428762436,
|
| 1577 |
+
"step": 196
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 1.256,
|
| 1581 |
+
"grad_norm": 3.893186569213867,
|
| 1582 |
+
"learning_rate": 0.001,
|
| 1583 |
+
"loss": 5.796,
|
| 1584 |
+
"mean_token_accuracy": 0.624050185084343,
|
| 1585 |
+
"step": 197
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 1.2624,
|
| 1589 |
+
"grad_norm": 7.488790512084961,
|
| 1590 |
+
"learning_rate": 0.001,
|
| 1591 |
+
"loss": 8.4914,
|
| 1592 |
+
"mean_token_accuracy": 0.5403816178441048,
|
| 1593 |
+
"step": 198
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 1.2688,
|
| 1597 |
+
"grad_norm": 10.15825080871582,
|
| 1598 |
+
"learning_rate": 0.001,
|
| 1599 |
+
"loss": 6.9975,
|
| 1600 |
+
"mean_token_accuracy": 0.5812895894050598,
|
| 1601 |
+
"step": 199
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 1.2752,
|
| 1605 |
+
"grad_norm": 5.484641075134277,
|
| 1606 |
+
"learning_rate": 0.001,
|
| 1607 |
+
"loss": 6.5434,
|
| 1608 |
+
"mean_token_accuracy": 0.6408251523971558,
|
| 1609 |
+
"step": 200
|
| 1610 |
+
},
|
| 1611 |
+
{
|
| 1612 |
+
"epoch": 1.2816,
|
| 1613 |
+
"grad_norm": 6.395635604858398,
|
| 1614 |
+
"learning_rate": 0.001,
|
| 1615 |
+
"loss": 7.7083,
|
| 1616 |
+
"mean_token_accuracy": 0.5760865062475204,
|
| 1617 |
+
"step": 201
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 1.288,
|
| 1621 |
+
"grad_norm": 7.382782936096191,
|
| 1622 |
+
"learning_rate": 0.001,
|
| 1623 |
+
"loss": 7.147,
|
| 1624 |
+
"mean_token_accuracy": 0.5934395641088486,
|
| 1625 |
+
"step": 202
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 1.2944,
|
| 1629 |
+
"grad_norm": 6.834714412689209,
|
| 1630 |
+
"learning_rate": 0.001,
|
| 1631 |
+
"loss": 7.6222,
|
| 1632 |
+
"mean_token_accuracy": 0.5739501044154167,
|
| 1633 |
+
"step": 203
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 1.3008,
|
| 1637 |
+
"grad_norm": 5.695088863372803,
|
| 1638 |
+
"learning_rate": 0.001,
|
| 1639 |
+
"loss": 7.475,
|
| 1640 |
+
"mean_token_accuracy": 0.5647561475634575,
|
| 1641 |
+
"step": 204
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 1.3072,
|
| 1645 |
+
"grad_norm": 5.658081531524658,
|
| 1646 |
+
"learning_rate": 0.001,
|
| 1647 |
+
"loss": 6.6691,
|
| 1648 |
+
"mean_token_accuracy": 0.6017946600914001,
|
| 1649 |
+
"step": 205
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 1.3136,
|
| 1653 |
+
"grad_norm": 9.058781623840332,
|
| 1654 |
+
"learning_rate": 0.001,
|
| 1655 |
+
"loss": 6.3451,
|
| 1656 |
+
"mean_token_accuracy": 0.6354028433561325,
|
| 1657 |
+
"step": 206
|
| 1658 |
+
},
|
| 1659 |
+
{
|
| 1660 |
+
"epoch": 1.32,
|
| 1661 |
+
"grad_norm": 6.336489200592041,
|
| 1662 |
+
"learning_rate": 0.001,
|
| 1663 |
+
"loss": 7.9945,
|
| 1664 |
+
"mean_token_accuracy": 0.5321162715554237,
|
| 1665 |
+
"step": 207
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 1.3264,
|
| 1669 |
+
"grad_norm": 14.819890022277832,
|
| 1670 |
+
"learning_rate": 0.001,
|
| 1671 |
+
"loss": 9.4337,
|
| 1672 |
+
"mean_token_accuracy": 0.5192963108420372,
|
| 1673 |
+
"step": 208
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 1.3328,
|
| 1677 |
+
"grad_norm": 6.626960754394531,
|
| 1678 |
+
"learning_rate": 0.001,
|
| 1679 |
+
"loss": 7.0263,
|
| 1680 |
+
"mean_token_accuracy": 0.6068531572818756,
|
| 1681 |
+
"step": 209
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 1.3392,
|
| 1685 |
+
"grad_norm": 7.5550384521484375,
|
| 1686 |
+
"learning_rate": 0.001,
|
| 1687 |
+
"loss": 7.5964,
|
| 1688 |
+
"mean_token_accuracy": 0.5845440551638603,
|
| 1689 |
+
"step": 210
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 1.3456000000000001,
|
| 1693 |
+
"grad_norm": 9.043644905090332,
|
| 1694 |
+
"learning_rate": 0.001,
|
| 1695 |
+
"loss": 8.9667,
|
| 1696 |
+
"mean_token_accuracy": 0.5143468156456947,
|
| 1697 |
+
"step": 211
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 1.3519999999999999,
|
| 1701 |
+
"grad_norm": 15.724257469177246,
|
| 1702 |
+
"learning_rate": 0.001,
|
| 1703 |
+
"loss": 7.97,
|
| 1704 |
+
"mean_token_accuracy": 0.5721839368343353,
|
| 1705 |
+
"step": 212
|
| 1706 |
+
},
|
| 1707 |
+
{
|
| 1708 |
+
"epoch": 1.3584,
|
| 1709 |
+
"grad_norm": 12.023893356323242,
|
| 1710 |
+
"learning_rate": 0.001,
|
| 1711 |
+
"loss": 7.454,
|
| 1712 |
+
"mean_token_accuracy": 0.599329836666584,
|
| 1713 |
+
"step": 213
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 1.3648,
|
| 1717 |
+
"grad_norm": 6.79151725769043,
|
| 1718 |
+
"learning_rate": 0.001,
|
| 1719 |
+
"loss": 7.7166,
|
| 1720 |
+
"mean_token_accuracy": 0.5553706511855125,
|
| 1721 |
+
"step": 214
|
| 1722 |
+
},
|
| 1723 |
+
{
|
| 1724 |
+
"epoch": 1.3712,
|
| 1725 |
+
"grad_norm": 6.837470054626465,
|
| 1726 |
+
"learning_rate": 0.001,
|
| 1727 |
+
"loss": 10.8317,
|
| 1728 |
+
"mean_token_accuracy": 0.4465588107705116,
|
| 1729 |
+
"step": 215
|
| 1730 |
+
},
|
| 1731 |
+
{
|
| 1732 |
+
"epoch": 1.3776,
|
| 1733 |
+
"grad_norm": 9.545637130737305,
|
| 1734 |
+
"learning_rate": 0.001,
|
| 1735 |
+
"loss": 6.7084,
|
| 1736 |
+
"mean_token_accuracy": 0.593764916062355,
|
| 1737 |
+
"step": 216
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 1.384,
|
| 1741 |
+
"grad_norm": 5.631826877593994,
|
| 1742 |
+
"learning_rate": 0.001,
|
| 1743 |
+
"loss": 6.8109,
|
| 1744 |
+
"mean_token_accuracy": 0.5985722541809082,
|
| 1745 |
+
"step": 217
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 1.3904,
|
| 1749 |
+
"grad_norm": 7.997573375701904,
|
| 1750 |
+
"learning_rate": 0.001,
|
| 1751 |
+
"loss": 7.6082,
|
| 1752 |
+
"mean_token_accuracy": 0.5720451474189758,
|
| 1753 |
+
"step": 218
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 1.3968,
|
| 1757 |
+
"grad_norm": 10.023303985595703,
|
| 1758 |
+
"learning_rate": 0.001,
|
| 1759 |
+
"loss": 8.4064,
|
| 1760 |
+
"mean_token_accuracy": 0.5365717485547066,
|
| 1761 |
+
"step": 219
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 1.4032,
|
| 1765 |
+
"grad_norm": 7.888483047485352,
|
| 1766 |
+
"learning_rate": 0.001,
|
| 1767 |
+
"loss": 8.3661,
|
| 1768 |
+
"mean_token_accuracy": 0.5367622226476669,
|
| 1769 |
+
"step": 220
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 1.4096,
|
| 1773 |
+
"grad_norm": 7.387235164642334,
|
| 1774 |
+
"learning_rate": 0.001,
|
| 1775 |
+
"loss": 7.0045,
|
| 1776 |
+
"mean_token_accuracy": 0.5931309461593628,
|
| 1777 |
+
"step": 221
|
| 1778 |
+
},
|
| 1779 |
+
{
|
| 1780 |
+
"epoch": 1.416,
|
| 1781 |
+
"grad_norm": 4.908820629119873,
|
| 1782 |
+
"learning_rate": 0.001,
|
| 1783 |
+
"loss": 7.457,
|
| 1784 |
+
"mean_token_accuracy": 0.5676573291420937,
|
| 1785 |
+
"step": 222
|
| 1786 |
+
},
|
| 1787 |
+
{
|
| 1788 |
+
"epoch": 1.4224,
|
| 1789 |
+
"grad_norm": 7.541878700256348,
|
| 1790 |
+
"learning_rate": 0.001,
|
| 1791 |
+
"loss": 8.5576,
|
| 1792 |
+
"mean_token_accuracy": 0.5507887750864029,
|
| 1793 |
+
"step": 223
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 1.4288,
|
| 1797 |
+
"grad_norm": 6.35562801361084,
|
| 1798 |
+
"learning_rate": 0.001,
|
| 1799 |
+
"loss": 7.1046,
|
| 1800 |
+
"mean_token_accuracy": 0.5813699215650558,
|
| 1801 |
+
"step": 224
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 1.4352,
|
| 1805 |
+
"grad_norm": 5.955767631530762,
|
| 1806 |
+
"learning_rate": 0.001,
|
| 1807 |
+
"loss": 7.5295,
|
| 1808 |
+
"mean_token_accuracy": 0.5491400882601738,
|
| 1809 |
+
"step": 225
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 1.4416,
|
| 1813 |
+
"grad_norm": 10.310933113098145,
|
| 1814 |
+
"learning_rate": 0.001,
|
| 1815 |
+
"loss": 7.4046,
|
| 1816 |
+
"mean_token_accuracy": 0.5788294672966003,
|
| 1817 |
+
"step": 226
|
| 1818 |
+
},
|
| 1819 |
+
{
|
| 1820 |
+
"epoch": 1.448,
|
| 1821 |
+
"grad_norm": 8.590178489685059,
|
| 1822 |
+
"learning_rate": 0.001,
|
| 1823 |
+
"loss": 6.6688,
|
| 1824 |
+
"mean_token_accuracy": 0.5879671573638916,
|
| 1825 |
+
"step": 227
|
| 1826 |
+
},
|
| 1827 |
+
{
|
| 1828 |
+
"epoch": 1.4544000000000001,
|
| 1829 |
+
"grad_norm": 7.05693244934082,
|
| 1830 |
+
"learning_rate": 0.001,
|
| 1831 |
+
"loss": 9.785,
|
| 1832 |
+
"mean_token_accuracy": 0.4796893522143364,
|
| 1833 |
+
"step": 228
|
| 1834 |
+
},
|
| 1835 |
+
{
|
| 1836 |
+
"epoch": 1.4607999999999999,
|
| 1837 |
+
"grad_norm": 7.907512187957764,
|
| 1838 |
+
"learning_rate": 0.001,
|
| 1839 |
+
"loss": 6.369,
|
| 1840 |
+
"mean_token_accuracy": 0.6156620383262634,
|
| 1841 |
+
"step": 229
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 1.4672,
|
| 1845 |
+
"grad_norm": 5.780457973480225,
|
| 1846 |
+
"learning_rate": 0.001,
|
| 1847 |
+
"loss": 7.7141,
|
| 1848 |
+
"mean_token_accuracy": 0.5557262748479843,
|
| 1849 |
+
"step": 230
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 1.4736,
|
| 1853 |
+
"grad_norm": 12.134108543395996,
|
| 1854 |
+
"learning_rate": 0.001,
|
| 1855 |
+
"loss": 8.9121,
|
| 1856 |
+
"mean_token_accuracy": 0.5156212523579597,
|
| 1857 |
+
"step": 231
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 1.48,
|
| 1861 |
+
"grad_norm": 6.636194229125977,
|
| 1862 |
+
"learning_rate": 0.001,
|
| 1863 |
+
"loss": 7.12,
|
| 1864 |
+
"mean_token_accuracy": 0.6036765873432159,
|
| 1865 |
+
"step": 232
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 1.4864,
|
| 1869 |
+
"grad_norm": 5.020055770874023,
|
| 1870 |
+
"learning_rate": 0.001,
|
| 1871 |
+
"loss": 6.3509,
|
| 1872 |
+
"mean_token_accuracy": 0.6170566231012344,
|
| 1873 |
+
"step": 233
|
| 1874 |
+
},
|
| 1875 |
+
{
|
| 1876 |
+
"epoch": 1.4928,
|
| 1877 |
+
"grad_norm": 10.214864730834961,
|
| 1878 |
+
"learning_rate": 0.001,
|
| 1879 |
+
"loss": 6.5362,
|
| 1880 |
+
"mean_token_accuracy": 0.6308254599571228,
|
| 1881 |
+
"step": 234
|
| 1882 |
+
},
|
| 1883 |
+
{
|
| 1884 |
+
"epoch": 1.4992,
|
| 1885 |
+
"grad_norm": 11.291118621826172,
|
| 1886 |
+
"learning_rate": 0.001,
|
| 1887 |
+
"loss": 8.2135,
|
| 1888 |
+
"mean_token_accuracy": 0.5482836216688156,
|
| 1889 |
+
"step": 235
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 1.5056,
|
| 1893 |
+
"grad_norm": 10.337629318237305,
|
| 1894 |
+
"learning_rate": 0.001,
|
| 1895 |
+
"loss": 7.4203,
|
| 1896 |
+
"mean_token_accuracy": 0.5811362043023109,
|
| 1897 |
+
"step": 236
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 1.512,
|
| 1901 |
+
"grad_norm": 8.213200569152832,
|
| 1902 |
+
"learning_rate": 0.001,
|
| 1903 |
+
"loss": 8.0874,
|
| 1904 |
+
"mean_token_accuracy": 0.5553788542747498,
|
| 1905 |
+
"step": 237
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 1.5184,
|
| 1909 |
+
"grad_norm": 35.60185623168945,
|
| 1910 |
+
"learning_rate": 0.001,
|
| 1911 |
+
"loss": 8.3935,
|
| 1912 |
+
"mean_token_accuracy": 0.5098748654127121,
|
| 1913 |
+
"step": 238
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 1.5248,
|
| 1917 |
+
"grad_norm": 10.800209045410156,
|
| 1918 |
+
"learning_rate": 0.001,
|
| 1919 |
+
"loss": 8.3645,
|
| 1920 |
+
"mean_token_accuracy": 0.5591541230678558,
|
| 1921 |
+
"step": 239
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 1.5312000000000001,
|
| 1925 |
+
"grad_norm": 10.682641983032227,
|
| 1926 |
+
"learning_rate": 0.001,
|
| 1927 |
+
"loss": 8.0174,
|
| 1928 |
+
"mean_token_accuracy": 0.5395123660564423,
|
| 1929 |
+
"step": 240
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 1.5375999999999999,
|
| 1933 |
+
"grad_norm": 16.261472702026367,
|
| 1934 |
+
"learning_rate": 0.001,
|
| 1935 |
+
"loss": 8.4602,
|
| 1936 |
+
"mean_token_accuracy": 0.5196909606456757,
|
| 1937 |
+
"step": 241
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 1.544,
|
| 1941 |
+
"grad_norm": 7.711510181427002,
|
| 1942 |
+
"learning_rate": 0.001,
|
| 1943 |
+
"loss": 7.8776,
|
| 1944 |
+
"mean_token_accuracy": 0.5560385212302208,
|
| 1945 |
+
"step": 242
|
| 1946 |
+
},
|
| 1947 |
+
{
|
| 1948 |
+
"epoch": 1.5504,
|
| 1949 |
+
"grad_norm": 10.960959434509277,
|
| 1950 |
+
"learning_rate": 0.001,
|
| 1951 |
+
"loss": 8.9444,
|
| 1952 |
+
"mean_token_accuracy": 0.5182236731052399,
|
| 1953 |
+
"step": 243
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 1.5568,
|
| 1957 |
+
"grad_norm": 7.446345806121826,
|
| 1958 |
+
"learning_rate": 0.001,
|
| 1959 |
+
"loss": 6.9821,
|
| 1960 |
+
"mean_token_accuracy": 0.6031645685434341,
|
| 1961 |
+
"step": 244
|
| 1962 |
+
},
|
| 1963 |
+
{
|
| 1964 |
+
"epoch": 1.5632000000000001,
|
| 1965 |
+
"grad_norm": 8.444716453552246,
|
| 1966 |
+
"learning_rate": 0.001,
|
| 1967 |
+
"loss": 7.3512,
|
| 1968 |
+
"mean_token_accuracy": 0.5782665312290192,
|
| 1969 |
+
"step": 245
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 1.5695999999999999,
|
| 1973 |
+
"grad_norm": 11.063419342041016,
|
| 1974 |
+
"learning_rate": 0.001,
|
| 1975 |
+
"loss": 8.5629,
|
| 1976 |
+
"mean_token_accuracy": 0.5185030549764633,
|
| 1977 |
+
"step": 246
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 1.576,
|
| 1981 |
+
"grad_norm": 8.18127727508545,
|
| 1982 |
+
"learning_rate": 0.001,
|
| 1983 |
+
"loss": 8.4838,
|
| 1984 |
+
"mean_token_accuracy": 0.5051439926028252,
|
| 1985 |
+
"step": 247
|
| 1986 |
+
},
|
| 1987 |
+
{
|
| 1988 |
+
"epoch": 1.5824,
|
| 1989 |
+
"grad_norm": 9.148998260498047,
|
| 1990 |
+
"learning_rate": 0.001,
|
| 1991 |
+
"loss": 8.8174,
|
| 1992 |
+
"mean_token_accuracy": 0.515631377696991,
|
| 1993 |
+
"step": 248
|
| 1994 |
+
},
|
| 1995 |
+
{
|
| 1996 |
+
"epoch": 1.5888,
|
| 1997 |
+
"grad_norm": 7.322795867919922,
|
| 1998 |
+
"learning_rate": 0.001,
|
| 1999 |
+
"loss": 7.4007,
|
| 2000 |
+
"mean_token_accuracy": 0.5654035359621048,
|
| 2001 |
+
"step": 249
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 1.5952,
|
| 2005 |
+
"grad_norm": 9.227678298950195,
|
| 2006 |
+
"learning_rate": 0.001,
|
| 2007 |
+
"loss": 9.3753,
|
| 2008 |
+
"mean_token_accuracy": 0.5103882774710655,
|
| 2009 |
+
"step": 250
|
| 2010 |
+
},
|
| 2011 |
+
{
|
| 2012 |
+
"epoch": 1.6016,
|
| 2013 |
+
"grad_norm": 13.938738822937012,
|
| 2014 |
+
"learning_rate": 0.001,
|
| 2015 |
+
"loss": 7.1796,
|
| 2016 |
+
"mean_token_accuracy": 0.5777264684438705,
|
| 2017 |
+
"step": 251
|
| 2018 |
+
},
|
| 2019 |
+
{
|
| 2020 |
+
"epoch": 1.608,
|
| 2021 |
+
"grad_norm": 23.90799903869629,
|
| 2022 |
+
"learning_rate": 0.001,
|
| 2023 |
+
"loss": 7.985,
|
| 2024 |
+
"mean_token_accuracy": 0.5455044433474541,
|
| 2025 |
+
"step": 252
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 1.6143999999999998,
|
| 2029 |
+
"grad_norm": 10.978887557983398,
|
| 2030 |
+
"learning_rate": 0.001,
|
| 2031 |
+
"loss": 7.6049,
|
| 2032 |
+
"mean_token_accuracy": 0.5702199786901474,
|
| 2033 |
+
"step": 253
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 1.6208,
|
| 2037 |
+
"grad_norm": 8.08303165435791,
|
| 2038 |
+
"learning_rate": 0.001,
|
| 2039 |
+
"loss": 7.5714,
|
| 2040 |
+
"mean_token_accuracy": 0.5959209054708481,
|
| 2041 |
+
"step": 254
|
| 2042 |
+
},
|
| 2043 |
+
{
|
| 2044 |
+
"epoch": 1.6272,
|
| 2045 |
+
"grad_norm": 9.555181503295898,
|
| 2046 |
+
"learning_rate": 0.001,
|
| 2047 |
+
"loss": 7.6343,
|
| 2048 |
+
"mean_token_accuracy": 0.5681372657418251,
|
| 2049 |
+
"step": 255
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 1.6336,
|
| 2053 |
+
"grad_norm": 8.243929862976074,
|
| 2054 |
+
"learning_rate": 0.001,
|
| 2055 |
+
"loss": 8.2715,
|
| 2056 |
+
"mean_token_accuracy": 0.5482152029871941,
|
| 2057 |
+
"step": 256
|
| 2058 |
+
},
|
| 2059 |
+
{
|
| 2060 |
+
"epoch": 1.6400000000000001,
|
| 2061 |
+
"grad_norm": 8.662081718444824,
|
| 2062 |
+
"learning_rate": 0.001,
|
| 2063 |
+
"loss": 7.8241,
|
| 2064 |
+
"mean_token_accuracy": 0.5627938807010651,
|
| 2065 |
+
"step": 257
|
| 2066 |
+
},
|
| 2067 |
+
{
|
| 2068 |
+
"epoch": 1.6463999999999999,
|
| 2069 |
+
"grad_norm": 8.32907772064209,
|
| 2070 |
+
"learning_rate": 0.001,
|
| 2071 |
+
"loss": 7.335,
|
| 2072 |
+
"mean_token_accuracy": 0.5363176316022873,
|
| 2073 |
+
"step": 258
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 1.6528,
|
| 2077 |
+
"grad_norm": 11.057466506958008,
|
| 2078 |
+
"learning_rate": 0.001,
|
| 2079 |
+
"loss": 9.5491,
|
| 2080 |
+
"mean_token_accuracy": 0.5060033053159714,
|
| 2081 |
+
"step": 259
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 1.6592,
|
| 2085 |
+
"grad_norm": 83.32975006103516,
|
| 2086 |
+
"learning_rate": 0.001,
|
| 2087 |
+
"loss": 9.0589,
|
| 2088 |
+
"mean_token_accuracy": 0.50930255651474,
|
| 2089 |
+
"step": 260
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 1.6656,
|
| 2093 |
+
"grad_norm": 8.165995597839355,
|
| 2094 |
+
"learning_rate": 0.001,
|
| 2095 |
+
"loss": 7.9928,
|
| 2096 |
+
"mean_token_accuracy": 0.563197672367096,
|
| 2097 |
+
"step": 261
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 1.6720000000000002,
|
| 2101 |
+
"grad_norm": 16.992876052856445,
|
| 2102 |
+
"learning_rate": 0.001,
|
| 2103 |
+
"loss": 9.8182,
|
| 2104 |
+
"mean_token_accuracy": 0.48126567155122757,
|
| 2105 |
+
"step": 262
|
| 2106 |
+
},
|
| 2107 |
+
{
|
| 2108 |
+
"epoch": 1.6784,
|
| 2109 |
+
"grad_norm": 8.095207214355469,
|
| 2110 |
+
"learning_rate": 0.001,
|
| 2111 |
+
"loss": 9.419,
|
| 2112 |
+
"mean_token_accuracy": 0.500243678689003,
|
| 2113 |
+
"step": 263
|
| 2114 |
+
},
|
| 2115 |
+
{
|
| 2116 |
+
"epoch": 1.6848,
|
| 2117 |
+
"grad_norm": 16.054994583129883,
|
| 2118 |
+
"learning_rate": 0.001,
|
| 2119 |
+
"loss": 8.7803,
|
| 2120 |
+
"mean_token_accuracy": 0.5126448944211006,
|
| 2121 |
+
"step": 264
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 1.6912,
|
| 2125 |
+
"grad_norm": 16.552730560302734,
|
| 2126 |
+
"learning_rate": 0.001,
|
| 2127 |
+
"loss": 8.9847,
|
| 2128 |
+
"mean_token_accuracy": 0.5172990635037422,
|
| 2129 |
+
"step": 265
|
| 2130 |
+
},
|
| 2131 |
+
{
|
| 2132 |
+
"epoch": 1.6976,
|
| 2133 |
+
"grad_norm": 13.06109619140625,
|
| 2134 |
+
"learning_rate": 0.001,
|
| 2135 |
+
"loss": 10.1881,
|
| 2136 |
+
"mean_token_accuracy": 0.49380549043416977,
|
| 2137 |
+
"step": 266
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 1.704,
|
| 2141 |
+
"grad_norm": 10.187677383422852,
|
| 2142 |
+
"learning_rate": 0.001,
|
| 2143 |
+
"loss": 8.4531,
|
| 2144 |
+
"mean_token_accuracy": 0.5305923670530319,
|
| 2145 |
+
"step": 267
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 1.7104,
|
| 2149 |
+
"grad_norm": 8.00294017791748,
|
| 2150 |
+
"learning_rate": 0.001,
|
| 2151 |
+
"loss": 8.55,
|
| 2152 |
+
"mean_token_accuracy": 0.5253442898392677,
|
| 2153 |
+
"step": 268
|
| 2154 |
+
},
|
| 2155 |
+
{
|
| 2156 |
+
"epoch": 1.7168,
|
| 2157 |
+
"grad_norm": 8.169543266296387,
|
| 2158 |
+
"learning_rate": 0.001,
|
| 2159 |
+
"loss": 10.1152,
|
| 2160 |
+
"mean_token_accuracy": 0.4511877968907356,
|
| 2161 |
+
"step": 269
|
| 2162 |
+
},
|
| 2163 |
+
{
|
| 2164 |
+
"epoch": 1.7231999999999998,
|
| 2165 |
+
"grad_norm": 8.316620826721191,
|
| 2166 |
+
"learning_rate": 0.001,
|
| 2167 |
+
"loss": 7.8197,
|
| 2168 |
+
"mean_token_accuracy": 0.5667182505130768,
|
| 2169 |
+
"step": 270
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 1.7296,
|
| 2173 |
+
"grad_norm": 7.95322847366333,
|
| 2174 |
+
"learning_rate": 0.001,
|
| 2175 |
+
"loss": 8.585,
|
| 2176 |
+
"mean_token_accuracy": 0.5186630189418793,
|
| 2177 |
+
"step": 271
|
| 2178 |
+
},
|
| 2179 |
+
{
|
| 2180 |
+
"epoch": 1.736,
|
| 2181 |
+
"grad_norm": 10.643982887268066,
|
| 2182 |
+
"learning_rate": 0.001,
|
| 2183 |
+
"loss": 8.1634,
|
| 2184 |
+
"mean_token_accuracy": 0.5534348115324974,
|
| 2185 |
+
"step": 272
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"epoch": 1.7424,
|
| 2189 |
+
"grad_norm": 11.432665824890137,
|
| 2190 |
+
"learning_rate": 0.001,
|
| 2191 |
+
"loss": 9.9191,
|
| 2192 |
+
"mean_token_accuracy": 0.4892067015171051,
|
| 2193 |
+
"step": 273
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 1.7488000000000001,
|
| 2197 |
+
"grad_norm": 12.11987590789795,
|
| 2198 |
+
"learning_rate": 0.001,
|
| 2199 |
+
"loss": 7.9962,
|
| 2200 |
+
"mean_token_accuracy": 0.5544887632131577,
|
| 2201 |
+
"step": 274
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 1.7551999999999999,
|
| 2205 |
+
"grad_norm": 16.141345977783203,
|
| 2206 |
+
"learning_rate": 0.001,
|
| 2207 |
+
"loss": 9.2237,
|
| 2208 |
+
"mean_token_accuracy": 0.5020880922675133,
|
| 2209 |
+
"step": 275
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 1.7616,
|
| 2213 |
+
"grad_norm": 11.97499942779541,
|
| 2214 |
+
"learning_rate": 0.001,
|
| 2215 |
+
"loss": 9.4026,
|
| 2216 |
+
"mean_token_accuracy": 0.4923320710659027,
|
| 2217 |
+
"step": 276
|
| 2218 |
+
},
|
| 2219 |
+
{
|
| 2220 |
+
"epoch": 1.768,
|
| 2221 |
+
"grad_norm": 20.49795913696289,
|
| 2222 |
+
"learning_rate": 0.001,
|
| 2223 |
+
"loss": 8.6141,
|
| 2224 |
+
"mean_token_accuracy": 0.5215461701154709,
|
| 2225 |
+
"step": 277
|
| 2226 |
+
},
|
| 2227 |
+
{
|
| 2228 |
+
"epoch": 1.7744,
|
| 2229 |
+
"grad_norm": 20.775779724121094,
|
| 2230 |
+
"learning_rate": 0.001,
|
| 2231 |
+
"loss": 11.6113,
|
| 2232 |
+
"mean_token_accuracy": 0.4201728478074074,
|
| 2233 |
+
"step": 278
|
| 2234 |
+
},
|
| 2235 |
+
{
|
| 2236 |
+
"epoch": 1.7808000000000002,
|
| 2237 |
+
"grad_norm": 8.977204322814941,
|
| 2238 |
+
"learning_rate": 0.001,
|
| 2239 |
+
"loss": 8.8081,
|
| 2240 |
+
"mean_token_accuracy": 0.49830809235572815,
|
| 2241 |
+
"step": 279
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 1.7872,
|
| 2245 |
+
"grad_norm": 11.948820114135742,
|
| 2246 |
+
"learning_rate": 0.001,
|
| 2247 |
+
"loss": 9.8821,
|
| 2248 |
+
"mean_token_accuracy": 0.49480871111154556,
|
| 2249 |
+
"step": 280
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 1.7936,
|
| 2253 |
+
"grad_norm": 10.269512176513672,
|
| 2254 |
+
"learning_rate": 0.001,
|
| 2255 |
+
"loss": 8.2235,
|
| 2256 |
+
"mean_token_accuracy": 0.5259484350681305,
|
| 2257 |
+
"step": 281
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 1.8,
|
| 2261 |
+
"grad_norm": 5.860231876373291,
|
| 2262 |
+
"learning_rate": 0.001,
|
| 2263 |
+
"loss": 8.3539,
|
| 2264 |
+
"mean_token_accuracy": 0.5288261473178864,
|
| 2265 |
+
"step": 282
|
| 2266 |
+
},
|
| 2267 |
+
{
|
| 2268 |
+
"epoch": 1.8064,
|
| 2269 |
+
"grad_norm": 9.967249870300293,
|
| 2270 |
+
"learning_rate": 0.001,
|
| 2271 |
+
"loss": 9.6113,
|
| 2272 |
+
"mean_token_accuracy": 0.45666269958019257,
|
| 2273 |
+
"step": 283
|
| 2274 |
+
},
|
| 2275 |
+
{
|
| 2276 |
+
"epoch": 1.8128,
|
| 2277 |
+
"grad_norm": 10.237951278686523,
|
| 2278 |
+
"learning_rate": 0.001,
|
| 2279 |
+
"loss": 7.8902,
|
| 2280 |
+
"mean_token_accuracy": 0.5173894092440605,
|
| 2281 |
+
"step": 284
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 1.8192,
|
| 2285 |
+
"grad_norm": 14.716999053955078,
|
| 2286 |
+
"learning_rate": 0.001,
|
| 2287 |
+
"loss": 9.5343,
|
| 2288 |
+
"mean_token_accuracy": 0.5037772506475449,
|
| 2289 |
+
"step": 285
|
| 2290 |
+
},
|
| 2291 |
+
{
|
| 2292 |
+
"epoch": 1.8256000000000001,
|
| 2293 |
+
"grad_norm": 10.649497032165527,
|
| 2294 |
+
"learning_rate": 0.001,
|
| 2295 |
+
"loss": 9.8439,
|
| 2296 |
+
"mean_token_accuracy": 0.48058757930994034,
|
| 2297 |
+
"step": 286
|
| 2298 |
+
},
|
| 2299 |
+
{
|
| 2300 |
+
"epoch": 1.8319999999999999,
|
| 2301 |
+
"grad_norm": 15.107233047485352,
|
| 2302 |
+
"learning_rate": 0.001,
|
| 2303 |
+
"loss": 8.1109,
|
| 2304 |
+
"mean_token_accuracy": 0.5341241806745529,
|
| 2305 |
+
"step": 287
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 1.8384,
|
| 2309 |
+
"grad_norm": 12.771748542785645,
|
| 2310 |
+
"learning_rate": 0.001,
|
| 2311 |
+
"loss": 9.2594,
|
| 2312 |
+
"mean_token_accuracy": 0.48890865594148636,
|
| 2313 |
+
"step": 288
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 1.8448,
|
| 2317 |
+
"grad_norm": 9.82298469543457,
|
| 2318 |
+
"learning_rate": 0.001,
|
| 2319 |
+
"loss": 8.4437,
|
| 2320 |
+
"mean_token_accuracy": 0.5308887958526611,
|
| 2321 |
+
"step": 289
|
| 2322 |
+
},
|
| 2323 |
+
{
|
| 2324 |
+
"epoch": 1.8512,
|
| 2325 |
+
"grad_norm": 22.638954162597656,
|
| 2326 |
+
"learning_rate": 0.001,
|
| 2327 |
+
"loss": 8.8215,
|
| 2328 |
+
"mean_token_accuracy": 0.5280868858098984,
|
| 2329 |
+
"step": 290
|
| 2330 |
+
},
|
| 2331 |
+
{
|
| 2332 |
+
"epoch": 1.8576000000000001,
|
| 2333 |
+
"grad_norm": 10.145191192626953,
|
| 2334 |
+
"learning_rate": 0.001,
|
| 2335 |
+
"loss": 9.1644,
|
| 2336 |
+
"mean_token_accuracy": 0.5151188969612122,
|
| 2337 |
+
"step": 291
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 1.8639999999999999,
|
| 2341 |
+
"grad_norm": 7.520384311676025,
|
| 2342 |
+
"learning_rate": 0.001,
|
| 2343 |
+
"loss": 8.3904,
|
| 2344 |
+
"mean_token_accuracy": 0.5109719932079315,
|
| 2345 |
+
"step": 292
|
| 2346 |
+
},
|
| 2347 |
+
{
|
| 2348 |
+
"epoch": 1.8704,
|
| 2349 |
+
"grad_norm": 16.988628387451172,
|
| 2350 |
+
"learning_rate": 0.001,
|
| 2351 |
+
"loss": 8.5715,
|
| 2352 |
+
"mean_token_accuracy": 0.5133798718452454,
|
| 2353 |
+
"step": 293
|
| 2354 |
+
},
|
| 2355 |
+
{
|
| 2356 |
+
"epoch": 1.8768,
|
| 2357 |
+
"grad_norm": 10.056252479553223,
|
| 2358 |
+
"learning_rate": 0.001,
|
| 2359 |
+
"loss": 8.0868,
|
| 2360 |
+
"mean_token_accuracy": 0.544953741133213,
|
| 2361 |
+
"step": 294
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 1.8832,
|
| 2365 |
+
"grad_norm": 27.852039337158203,
|
| 2366 |
+
"learning_rate": 0.001,
|
| 2367 |
+
"loss": 6.7302,
|
| 2368 |
+
"mean_token_accuracy": 0.594126746058464,
|
| 2369 |
+
"step": 295
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 1.8896,
|
| 2373 |
+
"grad_norm": 12.71786880493164,
|
| 2374 |
+
"learning_rate": 0.001,
|
| 2375 |
+
"loss": 9.0788,
|
| 2376 |
+
"mean_token_accuracy": 0.5247168093919754,
|
| 2377 |
+
"step": 296
|
| 2378 |
+
},
|
| 2379 |
+
{
|
| 2380 |
+
"epoch": 1.896,
|
| 2381 |
+
"grad_norm": 15.346375465393066,
|
| 2382 |
+
"learning_rate": 0.001,
|
| 2383 |
+
"loss": 8.3889,
|
| 2384 |
+
"mean_token_accuracy": 0.5260215997695923,
|
| 2385 |
+
"step": 297
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 1.9024,
|
| 2389 |
+
"grad_norm": 23.141864776611328,
|
| 2390 |
+
"learning_rate": 0.001,
|
| 2391 |
+
"loss": 8.985,
|
| 2392 |
+
"mean_token_accuracy": 0.5044504255056381,
|
| 2393 |
+
"step": 298
|
| 2394 |
+
},
|
| 2395 |
+
{
|
| 2396 |
+
"epoch": 1.9088,
|
| 2397 |
+
"grad_norm": 13.86075210571289,
|
| 2398 |
+
"learning_rate": 0.001,
|
| 2399 |
+
"loss": 9.9779,
|
| 2400 |
+
"mean_token_accuracy": 0.4863489866256714,
|
| 2401 |
+
"step": 299
|
| 2402 |
+
},
|
| 2403 |
+
{
|
| 2404 |
+
"epoch": 1.9152,
|
| 2405 |
+
"grad_norm": 12.302156448364258,
|
| 2406 |
+
"learning_rate": 0.001,
|
| 2407 |
+
"loss": 10.3654,
|
| 2408 |
+
"mean_token_accuracy": 0.4447909966111183,
|
| 2409 |
+
"step": 300
|
| 2410 |
+
}
|
| 2411 |
+
],
|
| 2412 |
+
"logging_steps": 1,
|
| 2413 |
+
"max_steps": 500,
|
| 2414 |
+
"num_input_tokens_seen": 0,
|
| 2415 |
+
"num_train_epochs": 4,
|
| 2416 |
+
"save_steps": 20,
|
| 2417 |
+
"stateful_callbacks": {
|
| 2418 |
+
"TrainerControl": {
|
| 2419 |
+
"args": {
|
| 2420 |
+
"should_epoch_stop": false,
|
| 2421 |
+
"should_evaluate": false,
|
| 2422 |
+
"should_log": false,
|
| 2423 |
+
"should_save": true,
|
| 2424 |
+
"should_training_stop": false
|
| 2425 |
+
},
|
| 2426 |
+
"attributes": {}
|
| 2427 |
+
}
|
| 2428 |
+
},
|
| 2429 |
+
"total_flos": 8.069164560285696e+16,
|
| 2430 |
+
"train_batch_size": 8,
|
| 2431 |
+
"trial_name": null,
|
| 2432 |
+
"trial_params": null
|
| 2433 |
+
}
|